Predicting Urban Dynamics with GPS data by Multi-Order Poisson Regression Model
Yanru Chen (Tokyo Tech) Hayakawa Yuta(Tokyo Tech), Tsubouchi Kota, Masamichi Shimosaka(Tokyo Tech)
第62回情報処理学会 ユビキタスコンピューティングシステム研究会 (IPSJ SIGUBI), 2019/6
機械学習 (Machine Learning) データサイエンス (Data Science)
- Forecasting people flow in urban regions (urban dynamics) is playing an increasingly important role in urban planning, emergency management, public services, and commercial activities. In this paper, we propose a Multi-Order Poisson Regression Model for urban dynamics prediction based on an enriched and generalized feature representation. In the proposed method, new features are produced by employing a variety of polynomial combinations of multiple factors which greatly affect people flow (e.g., time-of-the-day, day-of-the-week, weather situation, holiday information). The results obtained from an experiment with a massive GPS dataset show that the proposed method is capable of producing models which have higher prediction accuracy compared to the state-of-the-art method.
Predicting Urban Dynamics with GPS data by Multi-Order Poisson Regression Model（外部サイト／External Site Link）